As multi-cloud adoption grows, Databricks supports your journey to becoming a data-centric enterprise with a unified platform.

Databricks held its first offline 'Data + AI World Tour' in Korea on the 25th at the Intercontinental Harmony Bloom. At this event, Databricks introduced its latest solutions and shared successful domestic introduction cases that applied them.

“As a pioneer in data lakehouses, we are focused on making data and AI accessible to everyone,” said Jang Jeong-wook, CEO of Databricks Korea. “This event is a great opportunity to see Databricks’ product innovation trends firsthand,” he said.

“Databricks is like Apple, consolidating various workloads on a single platform to maximize efficiency,” said Chris D'Agostino, Global Field CTO of Databricks. “We will efficiently manage the massive data of companies, help make data-driven decisions, and ultimately achieve cost savings.”

■ Data Lakehouse, a powerful one-stop data platform

To become a data-centric organization, there are many considerations that need to be taken into account, such as establishing a single governance for data management to simplify multiple tasks, such as leveraging AI and ML, energy reduction, orchestration, and data synchronization.

Data Lakehouse is an open, integrated platform that supports data, analytics, and AI. It combines the flexibility, cost efficiency, and scalability of a data lake with the data management capabilities of a data warehouse to support 'business intelligence (BI)' and 'machine learning (ML)' for all data.

This is explained as being able to resolve complexity, secure openness through open source, and enable multi-cloud adoption. Warehouse does not support unstructured data, but Lakehouse adds the concept of a hyperscaler.

Safety and performance are ensured through integrated data management, while security and ease of use are ensured through integrated governance. “Ultimately, it’s the quality of the data coming in that determines quality,” said Chris D’Agostino, Global Field CTO.

Databricks also demonstrated superior performance compared to other companies through its strategy of developing and continuously updating the best Apache Spark version. It optimized disk storage and caching with 10 years of know-how and added the advantages of distributed computing technology.

■ E-Mart 24, Hanwha, Devsisters… Presenting customer data examples
Among the key customers who appeared as speakers, Lee Jae-kyung, CIO of E-Mart 24, said, “90% of retail companies have unstructured data, and 25% of sales losses occur due to incorrect operations. However, after introducing the Databricks platform, we were able to reduce costs through optimization during service development.”

E-Mart 24 has established a roadmap to build big data infrastructure, select tasks, and share and verify them with field workers. However, as the amount of data for providing various services has become increasingly large and diverse, and the need to build a rapid infrastructure to integrate and process scattered data has been felt, the Databricks platform has been applied, he explained.

For example, in the AI product recommendation service, the total analysis time was 27 hours and the total cost was 930,000 won, but through data and algorithm optimization and computing resource optimization, the total analysis time was reduced by 96% and 93% to 1 hour and the cost was reduced by 70,000 won, respectively.

In the manufacturing sector, Hanwha Hankisen DT Strategy Manager stepped forward.

Hanwha Corporation started out as a defense company, but as new businesses such as weapons chemicals, secondary batteries, and construction were added, a system that could respond in a timely manner to changes in the management environment was needed.

As companies with different business structures were integrated, siloed systems existed for each company, so after introducing the Databricks platform, flexible responses became possible, and a data-based decision-making system was established through data integration across business divisions.

In particular, we integrated SAP data of different versions and built a data analysis environment of various sizes required for DT. Hanwha said, “We expect to ultimately achieve DX by continuously integrating data through active use of AI and ML in the future.”

Databricks said that its future plans include ensuring that all policies and recommendations are followed when creating new data.
At the event, version 2.0 of 'Dolly', which was released last March, was introduced.

Only a few large companies can train huge AI models like ChatGPT, because as the models get bigger, the cost becomes astronomical along with the required GPUs.

Databricks expressed its ambition to implement ChatGPT’s functions without high-cost infrastructure through ‘Dolly’. Databricks claimed that it will provide an open-source-based model so that companies can train it themselves.

Meanwhile, at Databricks’ Data + AI event held on this day, Databricks Korea’s 1st anniversary customer awards ceremony was held. In the afternoon, data-centric corporate architecture was presented to customers under three themes: technology track, developer track, and customer case track. Speakers included Job Korea, Pinda, Musinsa, Gmarket, MS, and Megazone Cloud.